Shared bicycle borrowing and returning demand prediction method based on multi-source data, terminal and medium

Abstract

The invention discloses a shared bicycle borrowing and returning demand prediction method based on multi-source data, a terminal and a medium, and the method comprises the steps: obtaining shared bicycle riding order data, PO I data and historical weather data, and carrying out the statistics of the borrowing and returning amount of each parking site per hour for each parking site of a shared bicycle; time dimension features, space dimension features and weather dimension features of bicycle borrowing and returning demands of the shared bicycle stations are mined respectively, and the relation between the three dimension features and the bicycle borrowing and returning demands of the stations is determined; constructing interactive station passenger flow characteristics and similar station passenger flow characteristics; and constructing a bicycle borrowing and returning demand prediction model based on the interactive station passenger flow characteristics and the similar station passenger flow characteristics, and training the bicycle borrowing and returning demand model by using historical passenger flow data. According to the method, the relationship between various influence factors and the bicycle borrowing and returning demands of the shared bicycle parking station is deeply explored, bicycle borrowing and returning quantity prediction is realized, a reasonable bicycle allocation quantity is provided for shared bicycle scheduling, and the scheduling cost of a shared bicycle enterprise is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of shared bicycle management and scheduling, and in particular to a method, terminal and medium for predicting shared bicycle borrowing and returning demand based on multi-source data. Background Art

[0002] In recent years, shared bicycles have been deeply loved by urban residents for their green and low-carbon, flexible and convenient, low price, and the ability to achieve the "last mile" accessibility. They have gradually become an important part of urban residents' travel. On the one hand, the shared bicycle market has shown a good development trend in recent years, with the scale of shared bicycle users, annual riding volume, and average daily riding trips increasing year by year.

[0003] The temporal and spatial dynamic imbalance of demand refers to the large fluctuations in demand for shared bicycles over time and the uneven spatial distribution. As the scale of shared bicycle users and the number of rides increase year by year, the temporal and spatial dynamic imbalance of demand becomes more and more obvious, resulting in the frequent phenomenon of "no bicycles available and nowhere to return". This phenomenon not only affects the user experience, but also brings challenges to operating companies and urban management. Due to the temporal and spatial dynamic imbalance of demand, shared bicycle companies need to spend high operation and maintenance costs to balance the temporal and spatial mismatch of supply and demand.

[0004] At present, the main scheduling method adopted by shared bicycle operators is unified scheduling, that is, scheduling stations with a small or large number of bicycles in a concentrated period of time. This scheduling strategy has obvious lags, and no refined scheduling strategy is formulated in combination with the actual borrowing and returning needs of the parking stations. The fundamental solution to this problem is to accurately predict the short-term borrowing and returning demand at the station level, formulate the most cost-effective scheduling strategy, and achieve "pre-rainy day" advance scheduling and precise scheduling. Specifically, by accurately predicting the short-term borrowing volume of the station, if the borrowing demand of the station is large in the next period, the precise number of required bicycles will be put in place in time. Conversely, by accurately predicting the short-term return volume of the station, if the return demand of the station is large in the next period, the precise number of redundant bicycles will be taken away in time.

[0005] Therefore, it is necessary to accurately predict the short-term vehicle borrowing demand at the site and transform the original centralized scheduling strategy into demand-oriented precise scheduling to provide more refined and reliable data support for scheduling and deployment. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a method, terminal and medium for predicting the demand for borrowing and returning shared bicycles based on multi-source data in view of the above-mentioned defects of the prior art. The technical solution adopted by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for predicting demand for borrowing and returning shared bicycles based on multi-source data, wherein the method comprises:

[0008] Obtain shared bicycle riding order data, POI data, and historical weather data, and count the number of bicycles borrowed and returned at each shared bicycle parking station on an hourly basis;

[0009] The time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bikes at shared bicycle stations are respectively explored to determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bikes at the stations;

[0010] According to the relationship between the three-dimensional characteristics and the demand for borrowing and returning bicycles at the station, the passenger flow characteristics of interactive stations and similar stations are constructed;

[0011] A bicycle borrowing and returning demand prediction model is constructed based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and the bicycle borrowing and returning demand model is trained using historical passenger flow data. The bicycle borrowing and returning demand model is used to predict the bicycle borrowing and returning demand at shared bicycle stations.

[0012] In one implementation, the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bicycles at the shared bicycle station are respectively mined to determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bicycles at the station, including:

[0013] Based on the shared bicycle order data, determine the relationship between weekdays, weekends, holidays and hourly periods and the demand for borrowing and returning bicycles at the station from the time dimension;

[0014] Based on POI data, determine the relationship between the geographical location, surrounding facilities, population density, and road network density of shared bicycle parking stations and the demand for borrowing and returning bicycles at the stations from a spatial dimension;

[0015] Based on historical weather data, the relationship between temperature, precipitation, ground wind speed and the demand for borrowing and returning vehicles at the site is determined from the weather dimension.

[0016] In one implementation, based on the shared bicycle riding order data, the relationship between working days, weekends, holidays and hourly periods and the demand for borrowing and returning bicycles at the station is determined from the time dimension, including:

[0017] Through data visualization, we analyze the time-varying curves of the number of vehicles borrowed and returned at stations throughout the day on weekdays, weekends, and holidays, and identify the demand patterns for borrowing and returning vehicles at different time points;

[0018] Calculate the Spearman correlation coefficient between the passenger flow in this period and the passenger flow in the previous N periods;

[0019] Calculate the Spearman correlation coefficient between the passenger flow in this period and the passenger flow in the same period on different working days of the same week and the passenger flow in the same period on the same working day of different weeks.

[0020] In one implementation, based on POI data, the relationship between the geographical location, surrounding facilities, population density, and road network density of the shared bicycle parking station and the demand for borrowing and returning bicycles at the station is determined from a spatial dimension, including:

[0021] Analyze the heat map of car rental and return demand in different geographical locations during the same period through data visualization;

[0022] Based on correlation analysis, the correlation between the population size around the station, road network density, and the number of various POI facilities and the demand for borrowing and returning vehicles at the station is determined.

[0023] In one implementation, based on historical weather data, the relationship between temperature, precipitation, and ground wind speed and the demand for borrowing and returning vehicles at a site is determined from a weather dimension, including:

[0024] Through data visualization, we analyzed the scatter plots of different temperatures, ground wind speeds, and rainfall and the demand for borrowing and returning vehicles at the stations;

[0025] Based on correlation analysis, the strength of the correlation between temperature, ground wind speed and rainfall and the demand for borrowing and returning vehicles at the station is explored.

[0026] In one implementation, based on the relationship between the three-dimensional features and the demand for borrowing and returning vehicles at the site, passenger flow features of interactive sites and passenger flow features of similar sites are constructed, including:

[0027] Based on a number of shared bicycle riding order data, the OD pairs of stations with close interactions are screened out, and the most relevant influence time periods of the interactive stations are obtained based on correlation analysis, and the passenger flow characteristics of the interactive stations are constructed;

[0028] The pairwise passenger flow sequence similarity measurement method based on Euclidean distance is used to identify pairs of stations with similar passenger flow change trends and passenger flow quantities, and to construct passenger flow characteristics of similar stations.

[0029] In one implementation, a vehicle borrowing and returning demand prediction model is constructed based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and the vehicle borrowing and returning demand model is trained using historical passenger flow data, including:

[0030] The spatiotemporal graph convolutional network is used to model the passenger flow characteristics of interactive stations and similar stations, and to extract the temporal and spatial variation characteristics of the demand for borrowing and returning vehicles;

[0031] The time and space variation features output by the spatiotemporal graph convolutional network are input into the XGBoost model, and the high-dimensional features extracted from the weather, space and time features are used to construct a vehicle borrowing and returning demand prediction model, which is a combination model of the spatiotemporal graph convolutional network and the XGBoost model;

[0032] The vehicle borrowing and returning demand model is trained using historical passenger flow data, and grid search tuning and Bayesian optimization tuning are used to perform hyperparameter tuning on the spatiotemporal graph convolutional network and XGBoost model in turn.

[0033] In a second aspect, an embodiment of the present invention further provides a system for predicting demand for borrowing and returning shared bicycles based on multi-source data, wherein the system comprises:

[0034] The data statistics module is used to obtain the shared bicycle riding order data, POI data and historical weather data, and to count the number of borrowed and returned bicycles at each shared bicycle parking station on an hourly basis;

[0035] The relationship analysis module is used to mine the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bicycles at shared bicycle stations, and determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bicycles at the stations;

[0036] The feature construction module is used to construct the passenger flow characteristics of interactive stations and similar stations according to the relationship between the three-dimensional features and the demand for borrowing and returning bicycles at the stations;

[0037] The demand prediction module is used to build a bicycle borrowing and returning demand prediction model based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and use historical passenger flow data to train the bicycle borrowing and returning demand model, and the bicycle borrowing and returning demand model is used to predict the bicycle borrowing and returning demand at the shared bicycle station.

[0038] In the third aspect, an embodiment of the present invention further provides a terminal, wherein the terminal includes a memory, a processor, and a shared bicycle borrowing and returning demand prediction program based on multi-source data stored in the memory and executable on the processor, and when the processor executes the shared bicycle borrowing and returning demand prediction program based on multi-source data, the steps of the shared bicycle borrowing and returning demand prediction method based on multi-source data of any one of the above-mentioned schemes are implemented.

[0039] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein a program for predicting the demand for shared bicycle borrowing and returning based on multi-source data is stored on the computer-readable storage medium. When the program for predicting the demand for shared bicycle borrowing and returning based on multi-source data is executed by a processor, the steps of the method for predicting the demand for shared bicycle borrowing and returning based on multi-source data described in any one of the above-mentioned schemes are implemented.

[0040] Beneficial effects: Compared with the prior art, the present invention provides a method for predicting the demand for borrowing and returning shared bicycles based on multi-source data. The present invention first obtains shared bicycle riding order data, POI data and historical weather data, and counts the number of borrowing and returning bicycles at each parking site of shared bicycles on an hourly basis. Then, the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the borrowing and returning demand of shared bicycle sites are respectively mined to determine the relationship between the three-dimensional characteristics and the borrowing and returning demand of the site. Then, according to the relationship between the three-dimensional characteristics and the borrowing and returning demand of the site, the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites are constructed. Finally, a borrowing and returning demand prediction model is constructed based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and the borrowing and returning demand model is trained using historical passenger flow data. The borrowing and returning demand model is used to predict the borrowing and returning demand of shared bicycle sites. The present invention uses multi-source data such as shared bicycle riding order data, POI data, and historical weather data to deeply explore the relationship between various influencing factors and the demand for borrowing and returning bicycles at shared bicycle parking sites, realize the prediction of the borrowing and returning volume, and provide a reasonable number of bicycles for shared bicycle scheduling, which helps to reduce the scheduling costs of shared bicycle companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flowchart of a preferred embodiment of a method for predicting demand for shared bicycle borrowing and returning based on multi-source data provided in an embodiment of the present invention.

[0042] Figure 2 It is a flow chart of multi-source data acquisition and site hourly data statistics in one implementation of the present invention.

[0043] Figure 3 This is a heat map of the demand for car rental during the morning rush hour in various streets in Shenzhen in one implementation of the present invention.

[0044] Figure 4 It is a framework diagram of the STGCN_XGBoost combined prediction model in one implementation of the present invention.

[0045] Figure 5 This is a prediction effect diagram of the STGCN_XGBoost model in one implementation of the present invention.

[0046] Figure 6 It is a schematic diagram of the architecture of a shared bicycle borrowing and returning demand prediction system based on multi-source data provided by an embodiment of the present invention.

[0047] Figure 7 It is a principle block diagram of a terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations or steps, nor must they be executed in the order described. For example, some operations or steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0050] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0051] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, words such as "first" and "second" are used to distinguish between identical or similar items with substantially identical functions and effects. For example, the first control information and the second control information are only used to distinguish different control information, and their order is not limited.

[0052] Those skilled in the art can understand that the words "first", "second", etc. do not limit the quantity and execution order, and the words "first", "second", etc. do not necessarily limit the differences.

[0053] It should be further understood that the term “and / or” used in the present specification and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0054] In the existing shared bicycle operation, there are problems such as "no bikes available, nowhere to return" due to inaccurate demand forecasting. Traditional demand forecasting methods often rely on a single data source or a simple time series model, ignoring the complex factors that affect the demand for borrowing and returning shared bicycles, such as weather changes, spatial distribution, and passenger flow characteristics. Therefore, the present invention proposes a method for predicting the demand for borrowing and returning shared bicycles based on multi-source data, which uses multi-source data to improve the accuracy and applicability of the prediction. Specifically, this embodiment first obtains shared bicycle riding order data, POI data, and historical weather data, and counts the number of borrowed and returned bicycles at each parking site of the shared bicycle on an hourly basis. Then, the time dimension characteristics, spatial dimension characteristics, and weather dimension characteristics of the borrowing and returning demand of shared bicycle sites are respectively mined to determine the relationship between the three dimensional characteristics and the borrowing and returning demand of the site. Then, according to the relationship between the three dimensional characteristics and the borrowing and returning demand of the site, the passenger flow characteristics of the interactive site and the passenger flow characteristics of the similar site are constructed. Finally, a prediction model for the demand for borrowing and returning bicycles is constructed based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and the historical passenger flow data is used to train the borrowing and returning bicycle demand model, which is used to predict the demand for borrowing and returning bicycles at shared bicycle stations. The present invention uses multi-source data such as shared bicycle riding order data, POI data, and historical weather data to deeply explore the relationship between various influencing factors and the demand for borrowing and returning bicycles at shared bicycle parking sites, realize the prediction of the number of borrowing and returning bicycles, provide a reasonable number of bicycles to be deployed for shared bicycle scheduling, and help reduce the scheduling cost of shared bicycle companies.

[0055] The method for predicting the demand for shared bicycle borrowing and returning based on multi-source data of this embodiment can be applied to terminals, including intelligent product terminals such as mobile phones, smart TVs, and computers, such as Figure 1 As shown in , the method for predicting the demand for borrowing and returning shared bicycles based on multi-source data in this embodiment includes the following steps:

[0056] Step S100: Acquire shared bicycle riding order data, POI data, and historical weather data, and count the number of bicycles borrowed and returned at each shared bicycle parking site on an hourly basis.

[0057] In this embodiment, crawler technology can be used to obtain shared bicycle riding order data, POI (Point of Interest, which refers to point data in Internet electronic maps) data and historical weather data from various data open platforms. Among them, the shared bicycle riding order data includes riding order number, start riding time, latitude and longitude coordinates of the start riding location, end riding time, latitude and longitude coordinates of the end riding location and other attributes). POI data includes attributes such as name, address, coordinates, and category. Historical weather data includes attributes such as region, time, temperature, wind speed, humidity, and rainfall.

[0058] Next, the coordinates of the starting and ending points of the shared bicycle riding order data are separated, and then the Scikit-learn_NearestNeighbors function of the Python library is used to assign the borrowing and returning points to the nearest shared bicycle parking stations. Finally, the pandas_resample function is used to aggregate the time series data by hour and count the number of borrowing and returning bicycles at each station. Figure 2 As shown, this embodiment can read data by connecting to mongodb data to obtain riding order data, subway station in and out passenger flow data and weather data, and then determine the nearest matching parking site, and then perform statistics based on a 1h time granularity to count the number of borrowed and returned vehicles at each site.

[0059] Step S200: respectively mine the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bicycles at the shared bicycle station, and determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bicycles at the station.

[0060] In this embodiment, the acquired shared bicycle riding order data, POI data and historical weather data are firstly subjected to data preprocessing. After preprocessing, the above three dimensions and the site borrowing and returning bicycle demand are analyzed from the three dimensions of time, space and weather.

[0061] Specifically, this embodiment first determines the relationship between the demand for borrowing and returning vehicles at the station on weekdays, weekends, holidays and hourly periods based on the shared bicycle riding order data from the time dimension. First, through data visualization, the time change curve of the borrowing and returning volume of vehicles at the station throughout the day on weekdays, weekends and holidays is analyzed to identify the demand patterns for borrowing and returning vehicles at different time points, such as the commuting peak on weekdays or the demand for leisure vehicles on weekends. Then, based on the correlation analysis, the similarity characteristics of the passenger flow change trend are analyzed, specifically, the Spearman correlation coefficient of the passenger flow in this period and the passenger flow in the previous N periods is calculated; finally, based on the correlation analysis, the time periodic characteristics of the passenger flow are analyzed, specifically, the Spearman correlation coefficient of the passenger flow in this period is calculated with the passenger flow in the same period on different weekdays of the same week and the passenger flow in the same period on the same weekday of different weeks.

[0062] Furthermore, based on POI data, this embodiment determines the relationship between the geographical location of shared bicycle parking stations, surrounding facilities (such as subway stations, shopping malls, schools, etc.), population density, and road network density and the demand for borrowing and returning bicycles at the stations from a spatial dimension. In this embodiment, firstly, a heat map of the demand for borrowing and returning bicycles at different geographical locations during the same period is analyzed through data visualization, such as Figure 3 It is expressed as the thermal performance of the demand for borrowing cars on different streets during the morning peak hours. Secondly, based on the correlation analysis, the correlation between the population around the station, the road network density, and the number of various POI facilities and the demand for borrowing and returning cars at the station is determined.

[0063] Furthermore, this embodiment determines the relationship between temperature, precipitation, and ground wind speed and the demand for borrowing and returning vehicles at the station from the weather dimension based on historical weather data. This embodiment first analyzes the scatter plots of different temperatures, ground wind speeds, and rainfall and the demand for borrowing and returning vehicles at the station in turn through data visualization. Then, based on correlation analysis, the correlation strength of temperature, ground wind speed, and rainfall with the demand for borrowing and returning vehicles at the station is explored.

[0064] Step S300: construct the passenger flow characteristics of interactive stations and similar stations according to the relationship between the three-dimensional characteristics and the demand for borrowing and returning bicycles at the stations.

[0065] This embodiment can screen out the OD pairs of sites with close interactions based on a number of shared bicycle riding order data, and obtain the most relevant influence period of the interactive sites based on correlation analysis to construct the passenger flow characteristics of the interactive sites. Specifically, the OD data is counted through the shared bicycle riding order data. The average ratio of the number of times from site O to site D to the total number of bikes borrowed at site O within N days is the site OD interaction intensity, and the formula is as follows:

[0066]

[0067] Among them, S is the interaction strength between OD sites; is the number of times from site O to site D on day i; is the total number of borrowed vehicles at station O; N is the number of days;

[0068] The passenger flow characteristic formula of the interactive station at the same time is as follows:

[0069]

[0070] in, is the passenger flow of interactive site i at time t; is the interaction intensity between site i and site j at time t; is the number of cars borrowed at station j at time t.

[0071] Furthermore, a pairwise passenger flow sequence similarity measurement method based on Euclidean distance is used to identify pairs of stations with similar characteristics in passenger flow change trends and passenger flow quantities, and to construct passenger flow characteristics of similar stations.

[0072] Specifically, the Euclidean distance between the time series of the number of cars borrowed in one day at station i and station j is calculated, and the Euclidean distance threshold σ is set. If the similarity exceeds σ, the similarity is 0. The formula for calculating the station similarity is as follows:

[0073]

[0074] Among them, P ij is the similarity between site i and site j; S ijis the Euclidean distance between the passenger flow time series of station i and station j; σ is the Euclidean distance threshold, S min is the minimum Euclidean distance, S max is the maximum Euclidean distance.

[0075] The characteristic formula of similar sites at the same time is as follows:

[0076]

[0077] in, is the passenger flow of similar stations at station i at time t; P ij is the site ij similarity; is the number of cars borrowed at station j at time t.

[0078] Step S400: construct a bicycle borrowing and returning demand prediction model based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and use historical passenger flow data to train the bicycle borrowing and returning demand model, wherein the bicycle borrowing and returning demand model is used to predict the bicycle borrowing and returning demand at the shared bicycle station.

[0079] This embodiment uses a spatiotemporal graph convolutional network to model the passenger flow characteristics of interactive sites and similar sites, and extracts the temporal and spatial variation characteristics of the demand for borrowing and returning vehicles. Then, the temporal and spatial variation characteristics output by the spatiotemporal graph convolutional network are input into the XGBoost model, and the high-dimensional features extracted from the weather, spatial and temporal features are used to construct a forecasting model for the demand for borrowing and returning vehicles. The forecasting model for the demand for borrowing and returning vehicles is a combination model of the spatiotemporal graph convolutional network and the XGBoost model.

[0080] Specifically, this embodiment uses the Spatial-Temporal Graph Convolutional Network (STGCN) to model the spatiotemporal dependency between cycling stations. The spatial relationship between stations is learned through graph convolution (GCN), and the time series features are modeled in combination with the temporal convolutional network (TCN) to extract the characteristics of demand changes over time and space. Figure 4 As shown in the figure. Then, the spatiotemporal features output by the STGCN model (spatial-temporal graph convolutional network) are input into the XGBoost model to further enhance the ability to capture nonlinear features. The XGBoost model can effectively utilize high-dimensional features extracted from weather, spatial and temporal features, and make fine-grained predictions on the demand for borrowing and returning cars.

[0081] This embodiment uses 90 days of historical passenger flow data to train the obtained car borrowing and returning demand model. This embodiment uses 90 days of historical passenger flow data as the training data set, which is divided into training set, validation set and test set according to the ratio of 70%, 15% and 15% respectively. In order to ensure the temporal continuity of the data and the generalization ability of the model, the sliding window method is selected to generate training samples. The processed spatiotemporal weather features are input into the model for model training. First, the spatiotemporal features are preliminarily learned in the STGCN model, and then the output results are input into the XGBoost model. The prediction accuracy is improved through this combined model.

[0082] Next, grid search tuning and Bayesian optimization tuning are used to tune the hyperparameters of the spatiotemporal graph convolutional network and XGBoost model in turn. Specifically, in the STGCN model, hyperparameters such as the number of convolution layers and the size of the convolution kernel are adjusted to optimize the extraction of spatiotemporal features. At the same time, in the XGBoost model, grid search or Bayesian optimization methods are used to adjust key parameters such as the learning rate and tree depth to improve the overall performance of the model. This embodiment can also perform feature screening and combination based on time, space, and weather characteristics to ensure that the features of the model input can effectively express changes in demand. The sensitivity of the model to demand fluctuations is improved by gradually eliminating redundant features and introducing new features (such as temperature, holidays, etc.). Evaluation indicators such as MAPE (mean absolute percentage error) and RMSE (root mean square error) are used to quantify the prediction accuracy of the model on the validation set and test set. By comparing the prediction performance under different parameter combinations, the model is further optimized to ensure the prediction effect of the car borrowing and returning demand prediction model on the test data, such as Figure 5 shown.

[0083] In summary, this embodiment first obtains shared bicycle riding order data, POI data and historical weather data, and counts the number of borrowing and returning bicycles at each parking site of shared bicycles by hour. Then, the time dimension characteristics, spatial dimension characteristics and weather dimension characteristics of the borrowing and returning demand of shared bicycle sites are respectively mined to determine the relationship between the three dimensional characteristics and the borrowing and returning demand of sites. Then, according to the relationship between the three dimensional characteristics and the borrowing and returning demand of sites, the passenger flow characteristics of interactive sites and the passenger flow characteristics of similar sites are constructed. Finally, a borrowing and returning demand prediction model is constructed based on the passenger flow characteristics of the interactive sites and the passenger flow characteristics of similar sites, and the borrowing and returning demand model is trained using historical passenger flow data. The borrowing and returning demand model is used to predict the borrowing and returning demand of shared bicycle sites. This embodiment uses multi-source data such as shared bicycle riding order data, POI data and historical weather data to deeply explore the relationship between various influencing factors and the borrowing and returning demand of shared bicycle parking sites, realize the prediction of borrowing and returning vehicles, provide a reasonable number of bicycles to be deployed for shared bicycle scheduling, and help reduce the scheduling cost of shared bicycle companies.

[0084] Based on the above embodiments, the present invention also provides a system for predicting demand for shared bicycle borrowing and returning based on multi-source data. Figure 6 As shown in , the system includes: a data statistics module 10, a relationship analysis module 20, a feature construction module 30 and a demand prediction module 40. Specifically, the data statistics module is used to obtain shared bicycle riding order data, POI data and historical weather data, and to count the number of borrowing and returning bicycles at each shared bicycle parking site on an hourly basis. The relationship analysis module 20 is used to respectively mine the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the borrowing and returning demand of shared bicycle sites, and determine the relationship between the three-dimensional characteristics and the borrowing and returning demand of the site. The feature construction module 30 is used to construct the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites according to the relationship between the three-dimensional characteristics and the borrowing and returning demand of the site. The demand prediction module 40 is used to construct a borrowing and returning demand prediction model based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of the similar site, and use the historical passenger flow data to train the borrowing and returning demand model, and the borrowing and returning demand model is used to predict the borrowing and returning demand of shared bicycle sites.

[0085] The working principles of each module in the shared bicycle borrowing and returning demand prediction system based on multi-source data in this embodiment are the same as the principles of each step in the above method embodiment, and will not be repeated here.

[0086] Each module in the above-mentioned multi-source data-based shared bicycle rental and return demand prediction system can be implemented in whole or in part through software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the terminal in the form of hardware, or can be stored in the memory in the terminal in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0087] Based on the above embodiment, the present invention further provides a terminal, the principle block diagram of the terminal can be as follows: Figure 7 The terminal may include one or more processors 100 ( Figure 7 Only one is shown in the figure), a memory 101 and a computer program 102 stored in the memory 101 and executable on one or more processors 100, for example, a sleep analysis program based on multi-sensor data. When one or more processors 100 execute the computer program 102, the various steps in the embodiment of the sleep analysis method based on multi-sensor data can be implemented. Alternatively, when one or more processors 100 execute the computer program 102, the functions of various modules / units in the embodiment of the sleep analysis system based on multi-sensor data can be implemented, which is not limited here.

[0088] In one embodiment, the processor 100 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0089] In one embodiment, the memory 101 may be an internal storage unit of an electronic device, such as a hard disk or memory of the electronic device. The memory 101 may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Further, the memory 101 may also include both an internal storage unit of the electronic device and an external storage device. The memory 101 is used to store computer programs and other programs and data required by the terminal. The memory 101 may also be used to temporarily store data that has been output or is to be output.

[0090] Those skilled in the art will understand that Figure 7 The principle block diagram shown in the figure is only a block diagram of a partial structure related to the scheme of the present invention, and does not constitute a limitation on the terminal to which the scheme of the present invention is applied. The specific terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0091] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, operating database or other media used in the embodiments provided by the present invention may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double operational data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting demand for shared bicycle borrowing and returning based on multi-source data, characterized in that: The method comprises: Obtain shared bicycle riding order data, POI data, and historical weather data, and count the number of bicycles borrowed and returned at each shared bicycle parking station on an hourly basis; The time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bikes at shared bicycle stations are respectively explored to determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bikes at the stations; According to the relationship between the three-dimensional characteristics and the demand for borrowing and returning bicycles at the station, the passenger flow characteristics of interactive stations and similar stations are constructed; A bicycle borrowing and returning demand prediction model is constructed based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and the bicycle borrowing and returning demand model is trained using historical passenger flow data. The bicycle borrowing and returning demand model is used to predict the bicycle borrowing and returning demand at shared bicycle stations.

2. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 1 is characterized in that: The time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bikes at shared bicycle stations are respectively mined, and the relationship between the three dimensional characteristics and the demand for borrowing and returning bikes at the stations is determined, including: Based on the shared bicycle order data, determine the relationship between weekdays, weekends, holidays and hourly periods and the demand for borrowing and returning bicycles at the station from the time dimension; Based on POI data, determine the relationship between the geographical location, surrounding facilities, population density, and road network density of shared bicycle parking stations and the demand for borrowing and returning bicycles at the stations from a spatial dimension; Based on historical weather data, the relationship between temperature, precipitation, ground wind speed and the demand for borrowing and returning vehicles at the site is determined from the weather dimension.

3. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 2 is characterized in that: Based on the shared bicycle order data, the relationship between weekdays, weekends, holidays and hourly periods and the demand for borrowing and returning bicycles at the station is determined from the time dimension, including: Through data visualization, we analyze the time-varying curves of the number of vehicles borrowed and returned at stations throughout the day on weekdays, weekends, and holidays, and identify the demand patterns for borrowing and returning vehicles at different time points; Calculate the Spearman correlation coefficient between the passenger flow in this period and the passenger flow in the previous N periods; Calculate the Spearman correlation coefficient between the passenger flow in this period and the passenger flow in the same period on different working days of the same week and the passenger flow in the same period on the same working day of different weeks.

4. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 2 is characterized in that: Based on POI data, the relationship between the geographical location, surrounding facilities, population density, and road network density of shared bicycle parking stations and the demand for borrowing and returning bicycles at the stations is determined from a spatial dimension, including: Analyze the heat map of car rental and return demand in different geographical locations during the same period through data visualization; Based on correlation analysis, the correlation between the population size around the station, road network density, and the number of various POI facilities and the demand for borrowing and returning vehicles at the station is determined.

5. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 2 is characterized in that: Based on historical weather data, the relationship between temperature, precipitation, ground wind speed and the demand for borrowing and returning vehicles at the site is determined from the weather dimension, including: Through data visualization, we analyzed the scatter plots of different temperatures, ground wind speeds, and rainfall and the demand for borrowing and returning vehicles at the stations; Based on correlation analysis, the strength of the correlation between temperature, ground wind speed and rainfall and the demand for borrowing and returning vehicles at the station is explored.

6. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 1 is characterized in that: According to the relationship between the three-dimensional characteristics and the demand for borrowing and returning vehicles at the station, the passenger flow characteristics of interactive stations and similar stations are constructed, including: Based on a number of shared bicycle riding order data, the OD pairs of stations with close interactions are screened out, and the most relevant influence time periods of the interactive stations are obtained based on correlation analysis, and the passenger flow characteristics of the interactive stations are constructed; The pairwise passenger flow sequence similarity measurement method based on Euclidean distance is used to identify pairs of stations with similar passenger flow change trends and passenger flow quantities, and to construct passenger flow characteristics of similar stations.

7. The method for predicting demand for shared bicycle borrowing and returning based on multi-source data according to claim 1 is characterized in that: Building a vehicle borrowing and returning demand prediction model based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and using historical passenger flow data to train the vehicle borrowing and returning demand model, including: The spatiotemporal graph convolutional network is used to model the passenger flow characteristics of interactive stations and similar stations, and to extract the temporal and spatial variation characteristics of the demand for borrowing and returning vehicles; The time and space variation features output by the spatiotemporal graph convolutional network are input into the XGBoost model, and the high-dimensional features extracted from the weather, space and time features are used to construct a vehicle borrowing and returning demand prediction model, which is a combination model of the spatiotemporal graph convolutional network and the XGBoost model; The vehicle borrowing and returning demand model is trained using historical passenger flow data, and grid search tuning and Bayesian optimization tuning are used to perform hyperparameter tuning on the spatiotemporal graph convolutional network and XGBoost model in turn.

8. A bicycle-sharing rental and return demand prediction system based on multi-source data, characterized in that: The system comprises: The data statistics module is used to obtain the shared bicycle riding order data, POI data and historical weather data, and to count the number of borrowed and returned bicycles at each shared bicycle parking station on an hourly basis; The relationship analysis module is used to mine the time dimension characteristics, space dimension characteristics and weather dimension characteristics of the demand for borrowing and returning bicycles at shared bicycle stations, and determine the relationship between the three dimensional characteristics and the demand for borrowing and returning bicycles at the stations; The feature construction module is used to construct the passenger flow characteristics of interactive stations and similar stations according to the relationship between the three-dimensional features and the demand for borrowing and returning bicycles at the stations; The demand prediction module is used to build a bicycle borrowing and returning demand prediction model based on the passenger flow characteristics of the interactive site and the passenger flow characteristics of similar sites, and use historical passenger flow data to train the bicycle borrowing and returning demand model, and the bicycle borrowing and returning demand model is used to predict the bicycle borrowing and returning demand at the shared bicycle station.

9. A terminal, characterized in that: The terminal includes a memory, a processor, and a shared bicycle borrowing and returning demand prediction program based on multi-source data stored in the memory and executable on the processor. When the processor executes the shared bicycle borrowing and returning demand prediction program based on multi-source data, the steps of the shared bicycle borrowing and returning demand prediction method based on multi-source data as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program for predicting the demand for shared bicycle borrowing and returning based on multi-source data. When the program for predicting the demand for shared bicycle borrowing and returning based on multi-source data is executed by a processor, the steps of the method for predicting the demand for shared bicycle borrowing and returning based on multi-source data as described in any one of claims 1 to 7 are implemented.

Images